Why separate code, documentation, and telemetry
Infrastructure changes have administrative reach; a portal frontend change should not implicitly have that scope. Separate repositories make ownership and release evidence explicit. The infrastructure coordinator serializes host mutations even when builds originate in different repositories.
Failure domains
Local hosting makes the portal dependent on the host it describes. Cloud telemetry storage and independent wiki hosting preserve two recovery paths: evidence and instructions. A tunnel outage can prevent automated recovery; a LAN or console path remains necessary.
Adoption instead of replacement
Rebuilding a functioning host would introduce avoidable risks to model files, credentials, and workers. Adoption records the current state and applies bounded changes. A clean-machine reconstruction is a distinct workflow, not an excuse to overwrite the running server.
Architectural limits
The portal is an investigation surface, not a second production control plane. Existing NutsNews repositories retain ownership of their resources unless a reviewed migration explicitly transfers that ownership.